src/streamlit_app.py

import streamlit as st
import zipfile, io, os
from langchain_community.document_loaders import PyPDFLoader
from langchain.text_splitter import CharacterTextSplitter
from langchain.schema import Document
from langchain.embeddings import SentenceTransformerEmbeddings
from langchain.vectorstores import FAISS
from transformers import AutoTokenizer, AutoModelForQuestionAnswering, pipeline

# Cache the QA initialization so ingestion runs once per session
@st.cache_resource
def init_qa(zip_bytes):
    tmp_dir = "tmp_pdfs"
    # Clean up or create temp folder
    if os.path.exists(tmp_dir):
        for f in os.listdir(tmp_dir):
            os.remove(os.path.join(tmp_dir, f))
    else:
        os.makedirs(tmp_dir)

    # Extract uploaded ZIP
    with zipfile.ZipFile(io.BytesIO(zip_bytes)) as z:
        z.extractall(tmp_dir)

    # Load all PDFs
    docs = []
    for fname in os.listdir(tmp_dir):
        if fname.lower().endswith(".pdf"):
            loader = PyPDFLoader(os.path.join(tmp_dir, fname))
            docs.extend(loader.load())

    # Split into manageable chunks
    splitter = CharacterTextSplitter(chunk_size=1000, chunk_overlap=200)
    split_docs = splitter.split_documents(docs)

    # Build vector store
    embeddings = SentenceTransformerEmbeddings(model_name="all-MiniLM-L6-v2")
    vector_store = FAISS.from_documents(split_docs, embeddings)

    # Load the QA model and tokenizer
    tokenizer = AutoTokenizer.from_pretrained("deepset/roberta-base-squad2")
    model = AutoModelForQuestionAnswering.from_pretrained("deepset/roberta-base-squad2")
    qa_pipeline = pipeline("question-answering", model=model, tokenizer=tokenizer)

    return vector_store, qa_pipeline

# Streamlit UI
st.title("RoBERTa QA Streamlit App")
st.write("Upload a ZIP of PDFs to initialize the QA engine.")
zip_file = st.file_uploader("ZIP file", type=["zip"])

if zip_file:
    vector_store, qa = init_qa(zip_file.read())
    query = st.text_input("Ask a question:")
    if query:
        docs = vector_store.similarity_search(query, k=4)
        context = "\n\n".join([doc.page_content for doc in docs])
        # Run QA
        result = qa(question=query, context=context)
        answer = result.get("answer", "No answer found.")
        st.write(answer)
else:
    st.info("Awaiting ZIP upload.")